parameters.yaml

##
## This file describes the parameters for the run and will be read
## by the prepare_data.R script in /src
##

setup:
    data.location: https://embl.de/download/zeller/crc_meta/
    tag: 'g2_l75'
    ab.cutoff: 1e-03

# alpha
associations:
    alpha.meta: 1e-05
    alpha.single.study: 0.005
    mult.corr: 'fdr'
    pr.cutoff: 0.05
    log.n0: 1e-05
    log.n0.func: 1e-08

ref.studies:
    - FR-CRC
    - AT-CRC
    - CN-CRC
    - US-CRC
    - DE-CRC
ext.crc.studies:
    - IT-CRC
    - IT-CRC-2
    - JP-CRC
ext.studies:
    - CN-T2D
    - SE-T2D
    - DE-PD
    - US-HMP
    - DN-metaHIT

# model building
model.building:
    ab.cutoff: 1e-03
    norm.method: 'log.std'
    sd.min.q: 0.1
    log.n0: 1e-05
    log.n0.func: 1e-08
    n.p: 2
    norm.margin: 1
    num.folds: 10
    num.resample: 10
    stratify: TRUE
    ml.method: 'lasso'
    modsel.crit: 'pr'
    min.nonzero.coeff: 1
    feature.selection:
        type: "AUC"
        cutoff: 800


# Plotting
plotting:
    study.cols:
        FR-CRC: '#2FBFBF'
        AT-CRC: '#177254'
        CN-CRC: '#F2CC30'
        US-CRC: '#74B347'
        DE-CRC: '#8265CC'
    stage.cols:
        early: '#D8AE84'
        late: '#AE6C3D'
    loc.cols:
        LC: "#6BAED6"
        RC: "#DEEBF7"
        Rectum: "#08306B"
    sex.cols:
        F: '#FEB24C'
        M: '#FFFFB2'
    bmi.cols:
        lean: '#53D680'
        overweight/obese: '#086B27'
    age.cols:
        A: "#FF5959"
        D: "#AD0000"
    peformance.cols:
        - 'black'
        - 'darkgreen'
        - 'forestgreen'
        - 'chartreuse3'
        - 'lawngreen'
        - 'yellow'
    gmm.cols:
        amino acid degradation: "#F2AD00"
        organic acid metabolism: "#F98400"
        glycoprotein degradation: "brown"
        carbohydrate degradation: "#00A08A"
    group.cols:
        CTR: "#C2C2C2"
        CRC: '#d9544d'